Research On Single-channel Blind Separation Method Of Signals Overlapped In Time And Frequency Domain

Along with the growing complex electromagnetic environment and the fast development of wireless communication techniques is widely employed, the situation of a single antenna receiving multiple signals is also increasingly common. Consequently, blind single-channel separation of signals overlapped in the time and frequency domain or extraction of these source signals from the mixed signal is significant.In order to solve single-channel blind separation problem of UWB (Ultra Wide Band) wireless communication signals overlapped in the time and frequency domain, two algorithm are proposed in this paper and simulation experiments are carried out to confirm the effectiveness of the proposed algorithms. Simulation results demonstrate that achieves high precision and fast convergence speed and is not affected by initial value selection.1. Based on the signal characteristics, a signal model is set up.In the case of ignoring multipath fading, this blind separation problem can be converted to a multidimensional variable parameter estimation problem.2. A minimum value estimation algorithm based on genetic algorithm. According to the multi-peak value features of objective function, we can obtain the initial point of the minimum search algorithm with the global search of genetic algorithm. And then, we can estimate the various parameters of the multi-dimensional variables with the local search of the minimum search algorithm. At last, the separation of the mixed signal is completed by reconstructing the signal. Simulation results demonstrate that the algorithm can accurately estimate the various parameters, and can obtain the all source signals successfully.3. A method of a hybrid genetic-minimum search algorithm (HG-MSA) with a predictive estimation. Because of the model parameters P and tO estimated accuracy has important influence on the whole algorithm convergence, a to estimate presented here to improve the convergence speed of the algorithm. Based on the signal characteristics of the signal, a signal model is set up and a predictive estimation is carried on the time width modulation parameter P and the center parameter of time domain tO; we can obtain the initial point of the minimum search algorithm with the global search of genetic algorithm; we can estimate the various parameters of the multi-dimensional variables with the local search of the minimum search algorithm; In the end, we can reconstruct these source signals by the parameters of each sine modulated gauss pulse signal. Simulation results demonstrate that achieves high precision, fast convergence speed and is not affected by initial value selection.